Energy piles, which integrate ground heat exchangers into building foundations, represent a promising solution for exploiting shallow geothermal energy. In Brazil, where cooling demands are high, this technology has significant potential; however, its performance in hot-dominated climates may be limited by continuous cooling operation and progressive ground heat accumulation. This study experimentally investigates two types of energy piles, differing in material and geometry, installed as foundation elements in a living laboratory building in São Paulo, Brazil. The objective is to assess their short-term thermal performance, without direct comparison, and to provide guidance for cooling-oriented operation strategies. The piles were subjected to both continuous and intermittent heat rejection into the ground, with varying rest periods. Thermal sensors distributed along the pile depth were used to evaluate heat accumulation under elevated ground temperature conditions and to examine the influence of operating mode. Results from short-term performance tests indicate that intermittent operation enhances thermal efficiency for both pile types compared to continuous operation. This improvement is attributed to partial thermal recovery during rest periods. Furthermore, in the intermittent tests, particularly for the steel pile, no discernible effect of the pause duration between heating cycles was observed on the heat exchange rate per unit length of the active pile. Likewise, pause duration did not influence the temperature decay during off periods in pile segments embedded in strata exhibiting elevated groundwater flow velocities. • Two types of energy piles were tested under varying operation modes in a hot climate. • Short-term intermittent operation mitigates heat accumulation in energy piles. • Pause periods enhance short-term heat exchange and pile temperature recovery. • Soil stratification and groundwater flow influence subsurface heat dissipation. • The findings guide energy pile applications in tropical cooling-demand regions.
Sá et al. (Fri,) studied this question.